Recent markets have already shown multiple signs of a cooling in AI trading: AI-concept stocks have pulled back sharply from their highs. As of July 10, many AI-concept stocks that doubled in price within the year had retreated more than 20% from their intra-year peaks, with some individual stocks correcting by over 40%. In early July, technology stocks suffered sustained selling, with the Philadelphia Semiconductor Index plunging 6.27% in a single day.
Apollo Global Management Chief Economist Torsten Slok issued a warning on Monday that if the sell-off in artificial intelligence-related stocks intensifies further, the U.S. dollar’s recent rally will face significant risks. This alert reveals a widely overlooked structural vulnerability in current international capital flows.
Slok pointed out that over the past 12 months, international investors have actively positioned around the AI theme, driving record rolling net foreign inflows into U.S. equities. However, the vast majority of these investors have not hedged their currency exposure against exchange rate risk. This detail constitutes the core mechanism of the dollar’s “hidden dependence”: overseas investors need to convert their local currencies into U.S. dollars before purchasing U.S. stocks, which provides important support for the greenback. At the same time, compared with other major economies, U.S. interest rates remain elevated, making the cost of purchasing currency hedging tools prohibitively high and leaving the dollar more susceptible to equity market fluctuations. Slok warned that if AI disappoints and leads to a pullback in inflows, it would pose significant downside risk to the dollar. This implies that the dollar effectively “hides a dependence on AI trading.”
Deutsche Bank’s analysis echoes Slok’s assessment. Deutsche Bank Senior Financial Markets Analyst Mallika Sachdeva noted that global capital is flooding into U.S. tech stocks at an unprecedented scale, and this trend is fundamentally altering the dollar’s risk profile—it is gradually evolving from a traditional safe-haven asset into a “tech high-beta” asset highly correlated with the Nasdaq 100 Index. For a long time, the U.S. “twin deficits” have been primarily financed through foreign institutions purchasing U.S. Treasury bonds, a type of capital flow with clear countercyclical characteristics, where funds tend to flow simultaneously into Treasuries and the dollar during economic downturns, forming a natural cushioning mechanism. But Sachdeva warned that if the financing model shifts toward more cyclical equity-type flows, the dollar exchange rate will become riskier and more deeply dependent on AI-related market performance. Should AI profit expectations cool and U.S. stocks undergo a deep correction, foreign investors may simultaneously reduce equity holdings and actively hedge their dollar exposures, thereby undermining the dollar’s safe-haven support.
The scale of foreign inflows into U.S. equities in 2026 has indeed reached unprecedented levels. BofA’s strategy team cited EPFR data showing that for the week ending June 17, U.S. equity stock funds recorded net inflows of $119.2 billion, the highest in history. AI is the core driver of this flood of capital, as investors from Europe, Asia, and other regions believe their home markets lack AI-leading companies with sufficient scale and depth, leaving them no choice but to access this technology wave through U.S. stocks. Most overseas investors have not hedged currency risk; as they buy U.S. stocks, they must also actually buy U.S. dollars, further pushing up the dollar index. This makes the dollar’s strength increasingly dependent on equity inflows supported by the AI theme, rather than solely on traditional interest rate or trade factors.
Goldman Sachs (GS) Prime Brokerage reported that hedge funds are selling tech stocks at a record pace, with net exposure to the “Magnificent Seven” falling to its lowest level this year. BofA Securities (BAC) believes the biggest risk facing the market is that persistently surging oil prices force the Federal Reserve to maintain a hawkish stance, thereby triggering a deeper pullback in overvalued AI stocks.
Traders still expect that the Federal Reserve could raise rates as early as September. According to data from the prediction market Kalshi, as of July 13, traders priced in roughly a 54% probability that the Fed would raise rates by the end of this year. Slok’s warning uncovers a structural risk overlooked by the market: once AI trading recedes, the dollar may lose its most critical pillar of support.
